Most home computer users are now connected to a network such as the Internet in one way or another. The most popular connection technique still is to use the Public Switched Telephone Network (PSTN) and a device called a modem. As is now quite familiar to even the general population, a modem makes a connection by dialing a telephone number of an Internet Service Provider (ISP), who maintains equipment that connects to the Internet. Digital signals generated by the user's computer are converted to analog signals and vice versa by the modem such that they may be carried over the telephone lines accurately.
What is less familiar to the public at large is the configuration of the ISP equipment and how it provides connections to the Internet. ISPs such as America Online (AOL) maintain a very large number of dial-up access points. These access points permit a user to dial a local telephone number, which then connects the call to a local central office. The central office switch, which may be a so-called Class 5 switch, then directs the call to a dial termination point. The dial termination point may be located in or behind the central office, such as at a computer network Point of Presence (POP). At the POP, a device called a Remote Access Server (RAS) terminates the connection. There, Terminating Modems (TM) at the RAS are often aggregated together. In particular, the RAS contains a large number of modem devices that are used to connect to transmit and receive modem signals to and from the user Originating Modems (OM).
From the RAS, which converts signals back to digital form, the signals may be carried through a packet based network, such as an Internet Protocol (IP) network, to the Internet. In some instances, large service providers such as AOL contract with network service providers such as Genuity or UUNet to carry traffic from local central office switches to remote access server locations over high-speed digital lines.
However, other paradigms are resulting in fundamental changes in the nature of the telephone network. Most notably is the change to carry voice traffic from central offices over digital transport networks using technologies originally intended for carrying data traffic such as Internet Protocol (IP). So-called Voice-over IP (VoIP) packet networks are envisioned to be the architecture of choice of the future for voice transport.
In this architecture, shown at a high level in FIG. 1, a Central Office (CO) 12 can aggregate multiple Plain Old Telephone Service (POTS) type voice connections 10, multiplexing them into a digital Time Division Multiplex (TDM) transport 14 format such as T1 or E1 carriers. This permits the use of digital technologies to transport voice signals to a transit location in which is installed a Voice over Internet Protocol Gateway (VoIP GW) 20. The VoIP GW converts the TDM signals to a packet switched transport format, forwarding them to an IP network 30. At the other side of the IP network, a second VoIP GW 40 receives the signals, converts them back to TDM format, and forwards them to a far end Central Office (CO) 42 which then further forwards signal to individual far end POTS connections 44.
As telecom networks migrate to a VoIP architecture, it becomes important to support various types of calls that a user wishes to make over the TDM network. At present, there are standards for carrying voice, touchtone (Dual-Tone, Multi-Frequency (DTMF)) dialed digits, and fax signaling over IP connections. While there remains an effort to develop standards for carrying modem traffic over TDM connections, there is no standard yet adopted to date for reliable transport of modem signals over IP connections.
One effort towards solving this problem is so-called modem relay transport. Modem relay is being considered by the International Telecommunications Union (ITU) and Internet Engineering Task Force (IETF), with an aggressive schedule to ratify standards in the near future. The basic idea behind this architecture is to insert a “modem relay” capability into the VoIP GW. Such an architecture is shown in FIG. 2. Here the dial modem 14 acts as an origin point for a call to a destination point which may be an Internet Service Provider (ISP) 60. The modem call is first typically forwarded to a Class 5 or other central office switch in the standard fashion over a circuit switched PSTN 18. The Class 5 switch (not shown in FIG. 2) connects the call through the PSTN 18 to an Originating Voice Gateway (OGW) 20 which supports modem relay.
The OGW 20 implements some amount of modem intelligence so as to enable modem traffic to be carried over an IP network 20 to a Terminating Gateway (TGW) 40. This may consist, for example, of (de)modulating the modem data (e.g., per the CCITT V.90, V.34, V.32, or V.22 standards), error correction processing (e.g., per V.42 or V.44), and encapsulating the resulting data modem as a Simple Packet Relay Transport (SPRT) packet.
The TGW 40 receives this “Modem over IP” (MoIP) formatted packet and then converts it back to a TDM format so that it can be transported over another PSTN 44 connection to a Remote Access Server (RAS) 50. This involves stripping off the SPRT formatting, performing error correction V.42 and data modulation protocol (V.90, V.34, V.32, V.22) etc. formatting. From the Remote Access Server, the packet is then passed over a pure TDM network 44 to the ISP 60. Here, the data is (de)modulated and error corrected by the terminating modem (RAS).
In this modem relay architecture, both the OGW 20 and the TGW 40 must include some amount of modem intelligence in order to permit proper transport of the modem signals over the IP network. In particular, they should perform basic portions of a modem protocol stack processing, as shown. A Digital Signal Processor (DSP) located in each of the gateways 20 and 40 and at the RAS 50 performs the required protocol translations. At the lowest layer of the protocol stack, this includes a physical layer performing modulation/ demodulation or data “modem pump” functions in accordance with modem standards (V.90, V.34, V.32, V.22, and the like). The modem enabled gateways 20 and 40 also perform secondary physical layer functions such as error detection and error correction as specified by V.42 or V.44, for example.
The gateways 20 and 40 also perform tasks associated with network layer tasks. These may, for example, consist of layering a Simple Packet Relay Transport (SPRT) over UDP to format data signals so that they may be properly transported over the IP network 30. Note that the SPRT packets are still compressed (per V.42bis or V.44) when so forwarded.